Sharp regularity for singular obstacle problems

2025-05-03 0 0 310.55KB 29 页 10玖币
侵权投诉
arXiv:2210.09413v1 [math.AP] 17 Oct 2022
SHARP REGULARITY FOR SINGULAR OBSTACLE
PROBLEMS
DAMI ˜
AO J. ARA ´
UJO, RAFAYEL TEYMURAZYAN, AND VARDAN VOSKANYAN
Abstract. We obtain sharp local C1regularity of solutions for sin-
gular obstacle problems, Euler-Lagrange equation of which is given by
pu=γ(uϕ)γ1in {u > ϕ},
for 0 < γ < 1 and p2. At the free boundary {u > ϕ}, we prove
optimal C1regularity of solutions, with τgiven explicitly in terms of
p,γand smoothness of ϕ, which is new even in the linear setting.
MSC (2020): 35B65, 35J60, 35J75, 35B33, 49Q20, 49Q05.
Keywords: Singular obstacle problems, degenerate elliptic operators,
sharp regularity, free boundary.
1. Introduction
In this paper we study minimization problems with non-differentiable
zero order dependence. More precisely, in a bounded domain Rn, for a
constant γ(0,1), we study regularity of minimizers of
J(u) = inf
vKJ(v),(1.1)
where
J(v) := Z|∇v|p
p+ (vϕ)γdx,
and
K:= nvW1,p(Ω); vϕ, v gW1,p
0(Ω)o,(1.2)
with ϕC1(Ω), for a β(0,1] and gW1,p(Ω). The corresponding
Euler-Lagrange equation is
pu=γ(uϕ)γ1in {u > ϕ} ∩ ,(1.3)
where
pu:= div |∇u|p2u
is the standard p-Laplacian operator, 2 p < +. Note that the right
hand side in (1.3) blows up at the free boundary points
{u > ϕ} ∩ ,
which makes it essential to understand its effect on the regularity of mini-
mizers. The parameter γ, thus, measures the magnitude of singularity.
1
2 D.J. ARA ´
UJO, R. TEYMURAZYAN, AND V. VOSKANYAN
Problems like (1.1) are used, for example, to model the density of a certain
chemical in reaction with a porous catalyst pellet (see, for instance, [7]), and
due to their wide range of applications, were studied by many prominent
mathematicians. In the linear setting (p= 2), the extreme cases (γ= 0
and γ= 1) of (1.1) were studied in [2] and [9] with flat obstacles (ϕ0).
The case γ= 0 is related to jets flow and cavity problems, and minimizers
are known to have local Lipschitz (optimal) regularity, as is established by
Alt and Caffarelli in [2]. In the nonlinear setting the extreme case of γ= 0
was studied in [12], where Lipschitz regularity of minimizers is established.
These type of problems, often referred to as Bernoulli type problems, appear
in heat flows, [1], electrochemical machining, [17], etc. The case of γ= 1
resembles the classical obstacle problem, and its solution, as is shown by
Brezis and Kinderlehrer in [8], is of class C1,1,p= 2. In the nonlinear
setting, p > 2, the obstacle problem was studied in [3,13]. Its unique
solution, as is shown in [3], is of class C1
loc at the free boundary points with
α= min β, 1
p1.
The problem (1.1) was studied in [21] (with p= 2 and flat obstacle), where,
using a minimizer preserving scaling, it was shown that minimizers are lo-
cally of the class C1,γ
2γ, 0 < γ < 1 (see also [4] and [5], for the problem
governed by the infinity Laplacian and uniformly elliptic fully nonlinear op-
erators, respectively). There are, so called, monotonicity formulas available
in the linear case, which play a crucial role in the study of the problem.
For γ(0,1), the nonlinear case is covered in [18], where for obstacle type
problem (with zero obstacle) is proved that minimizers are locally of the
class C1with
α= min σ,γ
pγ,(1.4)
where σis the H¨older regularity exponent for the gradient of p-harmonic
functions (astands for any b < a). Actually, for n= 2, from [6] one
concludes that α=γ
pγ. Observe that in all the above results (except in [3])
the obstacle is assumed to be trivial, guaranteeing a vanishing gradient of
solutions at the free boundary points, which is essential in the analysis (in
[3] obstacle is assumed to be of the class C1, but the zero order dependence
of the functional is smooth). The methods, used to obtain those results, fail
to work in the presence of non-trivial obstacles with large gradients at the
free boundary, and a new approach is required to tackle the issue.
In this work, we prove sharp regularity for minimizers of (1.1) both locally
and at the free boundary points. More precisely, we show that minimizers
are locally of the class C1, where
α= min σ,γ
pγ,β
p1,(1.5)
SHARP REGULARITY FOR SINGULAR OBSTACLE PROBLEMS 3
and σ > 0 is the H¨older regularity exponent of the gradient of p-harmonic
functions. Note that our result extends the local regularity (1.4) of [18]
for problems with non-trivial obstacles, which is new even for the linear
case (p= 2). Moreover, at the free boundary points we obtain optimal C1
regularity for minimizers of (1.1), which, unlike local interior estimates, does
not depend on the regularity of p-harmonic functions. To be exact, we show
that minimizers of (1.1) at the free boundary are in C1, where
τ= min β, γ
pγ,(1.6)
which generalizes the optimal regularity result obtained in [21] for the linear
case and trivial obstacle. Thus, at free boundary points the interior regu-
larity result of [18] improves substantially. Indeed, as the obstacle in [18] is
assumed to be trivial, then from (1.6), we have
τ=γ
pγ,
which is better than αfrom (1.4). Observe also, that our result extends
(continuously) the optimal regularity result of [3], from smooth lower or-
der dependence to the singular setting (Theorem 6.2). Our approach is
r1+β
r1+τ
ϕ
u
{u > ϕ}
Figure 1. Detachment of ufrom ϕat the free boundary.
based on geometric tangential analysis and a fine perturbation combined
with adjusted scaling argument. Strictly speaking, we redeem regularity by
“tangentially accessing” the information available in the “flatness regime”
(for a rather comprehensive introduction to geometric tangential analysis,
we refer the reader to [23]). In the complementary case, i.e., when the gra-
dient of a solution is bounded from below at a free boundary point, we use
an “adjusted scaling” argument to ensure that in the limit we get a linear
elliptic equation without the zero order term. The idea of the adjustment is
to get rid of those terms that blow up at the limit.
The paper is organized as follows: in Section 2, we prove existence of mini-
mizers and in Section 3, establish local sharp C1regularity result (Theorem
4 D.J. ARA ´
UJO, R. TEYMURAZYAN, AND V. VOSKANYAN
3.1). In Section 4, we obtain optimal regularity at free boundary points for
minimizers with small gradient (Theorem 4.1). Section 5is devoted to the
adjusted scaling argument. Finally, in Section 6, we obtain sharp regularity
for minimizers with large gradient at the free boundary points (Theorem
6.1). We close the paper with two appendices, containing some auxiliary
technical results (Appendix A) and a list of several known ones (Appendix
B), that are used in the paper.
Notations and assumptions. Hereafter Br(x0) is the ball of radius r
centered at x0,Br(0) = Br, and |Br|stands for the volume of the ball
Br. When Ω = Brin (1.2), we will often use the notation Krinstead of K.
Additionally, for a given integrable function f, we denote by (f)rits average
on the ball of radius rcentered at the origin, i.e.,
(f)r:= 1
|Br|ZBr
f(x)dx.
To avoid repetition of arguments when applying the conclusions for different
set of functions, we introduce a function H:RnR+, which is assumed to
be of class C2and satisfy the following structural assumptions
|∇H(ξ)| ≤ Υω(|ξ|),
|D2H(ξ)| ≤ Λω(|ξ|)
|ξ|,
ηTD2H(ξ)ηλω(|ξ|)
|ξ||η|2,
(1.7)
with
ω(z) := κ1zp1+κ2z, z 0,
where ξ,ηRnand Υ >0, Λ λ > 0, κ10, κ20 are constants, and
κ1+κ2>0. We also will use the notation
G(z) := Zz
0
ω(ζ)=κ1
zp
p+κ2
z2
2, z R+.(1.8)
Remark 1.1. The function H(ξ) = p1|ξ|psatisfies the above conditions
with κ1= 1 and κ2= 0. The classical linear version, p= 2, is recovered by
assuming κ1= 0. An alternative example of a function Hsatisfying (1.7)
is constructed in Appendix A.
2. Existence of minimizers
In this section we show that there exists at least one minimizer of (1.1).
Unlike the regular case (γ=1), which is known to have a unique minimizer
(see, for example, [3,10,13,20,22]) in the singular setting this is not
assured.
SHARP REGULARITY FOR SINGULAR OBSTACLE PROBLEMS 5
Theorem 2.1. If ϕW1,p(Ω) L(Ω),gW1,p(Ω) and γ(0,1), then
there exists a minimizer uof (1.1). Moreover,
kukL(Ω) max kgkL(Ω),kϕkL(Ω).(2.1)
Proof. Set
m:= inf
vKJ(v)0.
If viKis a minimizing sequence, then for ii0,i0N, one has
0J(vi)m+ 1,
hence Z
|∇vi|pdx pJ(vi)p(m+ 1),
and the Poincar´e inequality yields that the sequence viis bounded in W1,p
0(Ω).
Therefore, by Rellich-Kondrachov theorem, there is a function uW1,p
0(Ω)
such that, up to a subsequence,
viuweakly in W1,p(Ω) and viuin Lp(Ω).
Notice that uK, and thus, using the lower semi-continuity of the Dirichlet
integral, we obtain
mJ(u)lim inf
j→∞ J(vi) = m,
i.e., uis a minimizer of (1.1).
To see (2.1), set
uM:= min{u, M} ∈ K,
where
M:= max kgkL(Ω),kϕkL(Ω).
Since uis a minimizer,
Z
|∇u|p
pdx Z
|∇uM|
pdx Z(uMϕ)γ(uϕ)γdx,
and thus
0Z{u>M}
|∇u|p
pdx Z(uMϕ)γ(uϕ)γdx 0,
which implies that u=uM. Therefore, −kϕkL(Ω) ϕuM.
3. Local C1regularity estimates
One of the main steps towards the optimal regularity of minimizers is
obtaining C1regularity for minimizers of
Jδ(v) := Z
(H(v) + δ(vϕ)γ)dx (3.1)
over the set K, defined by (1.2). Here δ[0,1], p2 and Hsatisfies (1.7).
The key step towards the regularity is the result on the decay of integral
摘要:

arXiv:2210.09413v1[math.AP]17Oct2022SHARPREGULARITYFORSINGULAROBSTACLEPROBLEMSDAMI˜AOJ.ARA´UJO,RAFAYELTEYMURAZYAN,ANDVARDANVOSKANYANAbstract.WeobtainsharplocalC1,αregularityofsolutionsforsin-gularobstacleproblems,Euler-Lagrangeequationofwhichisgivenby∆pu=γ(u−ϕ)γ−1in{u>ϕ},for0<γ<1andp≥2.Atthefreeboun...

展开>> 收起<<
Sharp regularity for singular obstacle problems.pdf

共29页,预览5页

还剩页未读, 继续阅读

声明:本站为文档C2C交易模式,即用户上传的文档直接被用户下载,本站只是中间服务平台,本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私,请立即通知玖贝云文库,我们立即给予删除!
分类:图书资源 价格:10玖币 属性:29 页 大小:310.55KB 格式:PDF 时间:2025-05-03

开通VIP享超值会员特权

  • 多端同步记录
  • 高速下载文档
  • 免费文档工具
  • 分享文档赚钱
  • 每日登录抽奖
  • 优质衍生服务
/ 29
客服
关注